Introduction

Acquired immunodeficiency syndrome (AIDS) is caused by the human immunodeficiency virus (HIV)1. Nearly 38.4 million people worldwide were infected by HIV in 2021, which remains a pandemic health issue [https://www.unaids.org/en (EB/OL)]. HIV-1 reverse transcriptase (RT) has important biochemical functions for viral replication as RNA/DNA-dependent DNA polymerase and ribonuclease H (RNase H). Thus, it has been considered as one of the most attractive targets for the treatment of AIDS2. Currently, RT inhibitors are mainly divided into nucleoside RT inhibitors (NRTIs) and non-nucleoside RT inhibitors (NNRTIs)3. As allosteric inhibitors, NNRTIs interfere with the normal function of RT by binding to the NNRTI binding pocket (NNIBP) that is located about 10 Å from the polymerase active site4. NNRTIs have been key components in highly active antiretroviral therapy (HAART) due to their promising anti-HIV-1 activities, high specificity, and relatively low toxicity5,6.

As shown in Fig. 1, Nevirapine (NVP, 1), efavirenz (EFV, 2), etravirine (ETR, 3), rilpivirine (RPV, 4) and doravirine (DOR, 5) are NNRTIs that have been approved by the U.S. FDA for the treatment of AIDS7. NNRTIs have a relatively low genetic barrier because of their allosteric binding, which leads to rapid emergence of drug-resistant strains during their clinical applications8. For example, the first-generation NNRTIs NVP and EFV, have shown dramatically reduced activities against the mutant strains K103N, Y181C and L100I9,10. However, the second generation of NNRTIs, including ETR, RPV and DOR, demonstrated promising activities against early NNRTI-resistant mutations. Nevertheless, some new resistant strains have been selected by the second generation of NNRTIs, such as K103N/Y181C and F227L/V106A for ETR, E138K and F227C for RPV, and V106A and F227L for DOR11,12. ETR and RPV (diarylpyrimidine, DAPY) suffer from poor solubility (ETR, ≪1 μg/mL at pH 7.0; RPV, 20 ng/mL at pH 7.0), which affects their pharmacokinetic (PK) properties13. Nevertheless, RPV and other DAPY compounds such as dapivirine (R147681) compensate for low solubility through formation of hydrophobic aggregates that improve their bioavailability14, a phenomenon also observed in other relevant cases15,16. In addition, ETR and RPV are inhibitors of cytochrome P450 (CYP) enzymes, which can decrease their effective dose, sometimes necessitating co-formulation with PK enhancers such as ritonavir to compensate17. Therefore, there is a pressing need to develop novel NNRTIs with improved anti-HIV-1 activities against resistant mutant strains and improved drug-like properties.

Fig. 1: Chemical structures of NNRTI drugs approved by U.S. FDA.
figure 1

Nevirapine (NVP, 1), efavirenz (EFV, 2), etravirine (ETR, 3), rilpivirine (RPV, 4) and doravirine (DOR, 5).

Full size image

The cocrystal structure of ETR/RT (PDB code: 3MEC, Fig. 2a) indicated that the aminobenzonitrile moiety (right-wing) acted on a rather plastic “groove”, namely tolerant region I of the HIV-1 NNIBP18,19. The tolerant region I, which is formed from V106, F227, L234, P236 and Y318, is a modifiable chemical space that can accommodate various substituents20. Taking ETR as the lead compound, our previous efforts have identified a variety of DAPY-typed NNRTIs by exploiting the tolerant region I19,20,21,22,23. Particularly, compound BH-11c (6) revealed remarkable anti-HIV-1 activities against the wild-type (WT) and several single mutant strains, including K103N, E138K and Y181C21. In addition, BH-11c demonstrated significantly enhanced water solubility (33.4 μg/mL at pH 7.0) and safety profiles compared to ETR. However, BH-11c did not show satisfactory activities toward F227L/V106A and K103N/Y181C double mutant HIV-1 strains. Hence, it is worth conducting systematical optimization based on BH-11c to yield novel NNRTIs with improved drug-resistance profiles.

Fig. 2: Design and optimization of DAPY-typed HIV-1 NNRTIs targeting the main chains of the residues in tolerant region I.
figure 2

a The cocrystal structure of ETR/RT (PDB code: 3MEC). b Predicted binding mode of BH-11c bound to the NNIBP (PDB code: 3MEC). c Predicted binding mode of newly designed representative compound 18a1 bound to the NNIBP (PDB code: 3MEC). The docking was performed using the Surflex-Dock of SYBYL-X 2.0, and the results were shown by PyMOL (http://www.pymol.org/). The distance between ligands and residues is indicated by dashed line. The bond angle is indicated by an arc dashed line.

Full size image

Particularly, the F227L/V106A mutation has gained attention due to its resistance to the current NNRTIs, as well as BH-11c11,21. A known effective medicinal chemistry strategy to overcome drug resistance is to establish interactions between the ligand and main chain atom(s) of the surrounding amino acid residues24,25. Significantly, there is an extended channel in the tolerant region I, consisting of F227 and P236, that leads to the solvent26,27,28. In order to better understand the interactions between BH-11c and the NNIBP of HIV-1 RT (PDB code: 3MEC)29, molecular docking studies were performed using the software Surflex-Dock SYBYL-X 2.030, and the results were shown in Fig. 2c by PyMOL (http://www.pymol.org/). In the presence of the methylene group, there is a bond angle of 114.2° between piperazinyl and phenyl rings in the right wing of BH-11c (Fig. 2b). This resulted in the sulfonyl group of BH-11c being 4–6.7 Å away from the main chains of F227, P236 and L234, which was too far to interact. Based on this analysis, we designed a series of HIV-1 inhibitors and the docking results of representative compound 18a1 are shown in Fig. 2c. We speculated that the removal of the methylene group could shorten these distances (~3 Å, Fig. 2c), which potentially form additional interactions with F227, P236 and L234. A nitrogen atom, cyano (CN) or trifluoromethyl group was introduced at position 3 of the right phenyl ring, with the aim of establishing additional interactions with V106. A cyano or cyanoethylene (CV) group was introduced as a privileged fragment in the left-wing. We postulated that such structural optimizations derived from BH-11c could generate additional interactions (especially backbone-binding interactions) with the residues in tolerant region I.

This research focuses on the design, synthesis, anti-HIV evaluation and preliminary structure-activity relationships (SARs) of DAPY-typed derivatives. Furthermore, the co-crystal structure determination and molecular dynamics (MD) simulation were utilized to investigate interaction modes between a representative compound and the binding pocket. Finally, drug-like properties, including water solubility, cytochrome P450 (CYP) inhibition, and PK properties were investigated in detail.

Results

Synthesis and characterization for the target compounds

As shown in Fig. 3, 1-Boc-piperazine was reacted with 2-bromo-5-nitropyridine, or 2-bromo-5-nitrobenzonitrile, or 1-bromo-4-nitro-2-(trifluoromethyl)benzene at 120 °C to obtain 8a–c, which were reduced under hydrogen atmosphere to yield intermediates 9a–c31,32. In Fig. 4, the intermediate 11 or 12 was obtained by the reaction of 2,4-dichloropyrimidine (10) with 4-hydroxy-3,5-dimethylbenzonitrile or (E)-3-(4-hydroxy-3,5-dimethylphenyl) acrylonitrile33. Next, 11 or 12 was treated with well-prepared compounds 9a–c to provide the corresponding key intermediates 13a–c or 14a–c via a Buchwald-Hartwig coupling reaction23. Subsequently, the Boc group of compounds 13a–c or 14a–c was removed in the presence of trifluoroacetic acid to offer compounds 15a–c or 16a–c34. Finally, the target compounds 17a(1–5), 17b(1–5), 17c(1–5), 18a(1–6), 18b(1–6), or 18c(1–6) were synthesized by reacting intermediates 15a–c or 16a–c with different sulfonyl chlorides or acryloyl chloride, respectively35.

Fig. 3: The synthetic route of 9a–c.
figure 3

(i) 1-Boc-piperazine, K2CO3, DMF, 120 °C; (ii) H2, Pd/C, r.t.

Full size image
Fig. 4: The synthetic route of 17a(1–5), 17b(1–5), 17c(1–6), 18a(1–5), 18b(1–6) and 18c(1–6).
figure 4

(i) 4-hydroxy-3,5-dimethylbenzonitrile or (E)-3-(4-hydroxy-3,5-dimethylphenyl)acrylonitrile, K2CO3, DMF, 50 °C; (ii) 9a–c, Pd(OAc)2, xantphos, Cs2CO3, N2, 1,4-dioxane, 90 °C; (iii) TFA, DCM, r.t.; (iv) different sulfonyl chlorides or acyl chloride, TEA, DCM, 0 °C.

Full size image

Anti-HIV activities and SAR analysis

The newly synthesized DAPY-typed derivatives were evaluated for their cytotoxicity and activities against HIV-1 WT (IIIB) and K103N/Y181C double mutant (RES056) strains, as well as HIV-2 (ROD) via the MTT method in MT-4 cells36. The lead compound BH-11c and the six approved drugs zidovudine (AZT), NVP, EFV, ETR, RVP and DOR were used as control drugs. The biological results were determined as EC50 (anti-HIV potency), CC50 (cytotoxicity) and SI (selectivity index, the ratio of CC50/EC50).

As shown in Table 1, the newly synthesized compounds displayed moderate to excellent antiretroviral activity against WT HIV-1 strain with EC50 values ranging from 0.65 μM to 0.0014 μM. Among them, compounds 18a1 (EC50 = 0.0018 μM) and 18b1 (EC50 = 0.0014 μM) were the most potent inhibitors, which are evidently better than six control compounds, i.e., BH-11c (EC50 = 0.0027 μM), ETR (EC50 = 0.0033 μM), EFV (EC50 = 0.0037 μM), AZT (EC50 = 0.032 μM), NVP (EC50 = 0.19 μM) and DOR (EC50 = 0.013 μM). In particular, 18b1 displayed antiviral activity similar to the most potent reference RPV (EC50 = 0.0010 μM, CC50 = 3.98 μM), with relatively low cytotoxicity (CC50 = 10.15 μM). In addition, seven compounds 18a(2–5) and 18b(2–4) showed extremely potent anti-HIV-1 activities with EC50 values ranging from 0.0049 μM to 0.0026 μM, which were equipotent to BH-11c, ETR and EFV, and more active than AZT, NVP and DOR. For RES056, two compounds 18b1 (EC50 = 0.093 μM) and 18b2 (EC50 = 0.089 μM) exhibited the best antiviral activities, which were markedly more potent than BH-11c (EC50 = 0.59 μM), EFV (EC50 = 0.30 μM) and NVP (EC50 > 15.02 μM), and were equipotent to ETR (EC50 = 0.076 μM). As expected, no compounds revealed anti-HIV-2 activities since DAPY-typed derivatives interact with the NNIBP of HIV-1 RT specifically (data were not shown).

Table 1 Antiretroviral activities against HIV-1 IIIB and RES056 strains, cytotoxicity and selectivity index of the target compounds.
Full size table

According to the results in Table 1, the preliminary SARs and structure-cytotoxicity relationships (SCRs) can be depicted as follow:

  1. 1.

    By comparing series 17 and 18, it is obvious that the cyanovinyl group at R1 position is more preferred than the cyano group for anti-HIV-1 activity. However, cyanovinyl-containing compounds exhibited enhanced cytotoxicity possibly due to the “Michael addition effect”.

  2. 2.

    Next, we turned our attention to substituents at position 3 of the right phenyl ring (X in scheme 2). By comparing the corresponding derivatives in a, b and c series, the potency of compounds against WT HIV-1 was in the following order: b series (X = C-CN) ≥ a series (X = N) ≥ c series (X = C-CF3). The cytotoxicity of corresponding compounds in a, b and c series was as follows: when R1 is cyano group, 17a series (X = N) ≥ 17c series (X = C-CF3) ≥ 17b series (X = C-CN), except compounds 17a2/17c2 and 17b6/17c6; when R1 is cyanovinyl group, 18a series (X = N) ≥ 18b series (X = C-CN) ≥ 18c series (X = C-CF3).

  3. 3.

    In the case of the terminal substituents on piperazinyl (R2), the order of potency in 17c series was as follow: 17c1 (R2 = methyl sulfonyl) ≥17c2 (R2 = ethyl sulfonyl) ≥17c5 (R2 = cyclopropane sulfonyl) ≥17c3 (R2 = 2-propane sulfonyl) ≥17c4 (R2 = N,N-dimethyl sulfonyl) ≥17c6 (R2 = acryloyl). The terminal substituents in other series have a similar effect on the antiviral activity. Moreover, it is apparent that compounds with an acryloyl substituent indicated higher cytotoxicity than other sulfonyl substituents, except compound 17c6.

According to the anti-HIV-1(WT) results, the promising compounds 18a(1–5), 18b(1–5) and 18c1 were chosen to further evaluate their inhibitory potency against the prevalent HIV-1 mutations, including L100I, K103N, Y181C, Y188L, E138K, and F227L/V106A. As shown in Table 2, compounds 18b(1–5) revealed excellent inhibitory effect on these HIV-1 strains. Among them, compound 18b1 which exhibited the most potent antiviral activity against these HIV-1 variants, apart from Y188L. The results could be summarized as follow:

  1. 1.

    The one of most drug resistant variants F227L/V106A: compounds 18b1 (EC50 = 0.0099 μM) and 18b2 (EC50 = 0.011 μM) demonstrated the best antiviral potency, which were ~ 20-fold more potent than lead compound BH-11c (EC50 = 0.21 μM), and better than the reference drugs (ETR, EC50 = 0.025 μM; EFV, EC50 = 0.26 μM; NVP, EC50 > 15.02 μM; RPV, EC50 = 0.082 μM). In addition, compounds 18b(3–5) and 18c1 displayed two-digit nanomolar antiviral efficacy equivalent to that of ETR.

  2. 2.

    The L100I mutant strain: compounds 18b(1–3) and 18b5 demonstrated single-digit nanomolar inhibitory effect, which were more potent than all control compounds except RPV (EC50 = 0.0015 μM). Furthermore, compounds 18b4 (EC50 = 0.020 μM) and 18c1 (EC50 = 0.015 μM) presented the similar potency to the lead compound BH-11c (EC50 = 0.014 μM) and ETR (EC50 = 0.012 μM).

  3. 3.

    The K103N strain: 18b1 (EC50 = 0.0017 μM) was substantially more active than the lead compound BH-11c (EC50 = 0.0037 μM) and ETR (EC50 = 0.0043 μM). Compounds 18b(2–5) also showed high inhibitory activities with the EC50 values ranging from 0.0062 μM to 0.0035 μM.

  4. 4.

    The Y181C strain: compounds 18b1 (EC50 = 0.0064 μM) and 18b2 (EC50 = 0.0063 μM) showed 2.8-fold potency enhancement over ETR (EC50 = 0.018 μM). The antiretroviral activities of 18b3 (EC50 = 0.012 μM) and 18b4 (EC50 = 0.016 μM) were superior to the lead compound BH-11c (EC50 = 0.027 μM).

  5. 5.

    The Y188L strain: compounds 18b(1–3) and 18b5 displayed submicromolar antiviral activities, which were more active than the lead compound BH-11c (EC50 = 1.32 μM), though they were still weaker than ETR (EC50 = 0.025 μM) and RPV (EC50 = 0.079 μM).

  6. 6.

    The E138K mutant strain: compounds 18b(1–3) and 18b5 showed better or equivalent antiviral activities compared to ETR (EC50 = 0.018 μM), with EC50 values between 0.0019 μM and 0.0064 μM. Moreover, 18b1 (EC50 = 0.0064 μM) was slightly better than BH-11c (EC50 = 0.0094 μM), and was equipotent to RPV (EC50 = 0.0058 μM).

Table 2 Antiretroviral activity of selected compounds against HIV-1 mutant strains.
Full size table

In summary, 18b1 demonstrated superior potency against these HIV-1 mutant strains over the lead compound BH-11c. Furthermore, 18b1 possessed single-digit nanomolar potency against five mutants including L100I, K103N, Y181C, E138K and F227L/V106A. Also, compound 18b1 is about 3-fold more potent than ETR.

WT HIV-1 RT inhibition assay

In order to confirm the drug target of the synthesized DAPY derivatives, representative compounds were selected to test their inhibitory activity against WT HIV-1 RT. As shown in Table 3, these selected compounds exhibited inhibitory effect on HIV-1 RT with IC50 values between 0.19 to 0.056 μM, being evidently better than RVP (IC50 = 0.43 μM) and ETR (IC50 = 1.35 μM). The results indicated that these compounds target HIV-1 RT, and they could be classified as HIV-1 NNRTIs.

Table 3 Inhibitory activity against WT HIV-1 RT.
Full size table

Crystal structure of HIV-1 RT in complex with 18b1

We have determined the crystal structure of HIV-1 wild-type (WT) RT in complex with 18b1 at 2.5 Å resolution (Supplementary Table S1 and Supplementary Fig. S1) to gain a thorough understanding of the previously explained SAR. In general, the structure is very similar to previously reported RT-RPV37 and RT-25a3 structures (Fig. 5a, Supplementary Fig. S1). Notably, though, the cyano group of 18b1 displays hydrophobic contacts with the side chains of V106 and F227, and the sulfonyl group forms hydrogen bonds with the main chains of F227 and L234 (Fig. 5b). The comparison with previous structures shows that the positioning of 18b1 in the NNIBP is similar to RPV, except for the piperazine sulfonyl moiety that protrudes into the tolerant region I—as in the RT-25a complex (PDB 6C0N)—provoking the uplift of the loop preceding β9 and the one connecting β10-β11 (Fig. 5c). However, the removal of the linking carbon present in 25a (alongside the previously mentioned contacts of the cyano group) accounts for a substantial shift in the piperazine sulfonyl moiety positioning in comparison to the piperidine-linked benzenesulfonamide of 25a. Comparison of the protein-ligand contacts (Supplementary Fig. S2) indicate that the main interactions are similar for 25a and 18b1. A thorough analysis (using http://www.ebi.ac.uk/thornton-srv/databases/cgi-bin/pdbsum/GetPage.pl?pdbcode=index.html) reveals that 18b1 has a larger number of non-bonded contacts, especially in terms of backbone interactions (see log files in the Supplementary Information).

Fig. 5: Crystal structure of HIV-1 RT in complex with 18b1 (PDB ID: 8FE8).
figure 5

a 18b1 binding in the NNIBP. Color legend: (i) RT p66 subdomains: palm in dark red, thumb in green; (ii) RT p51 subunit in silver. b 18b1 cyano and sulfonyl groups contacts with RT. c Comparison of the RT complexes with RPV (PDB 4G1Q) and 25a (PDB 6C0N). d Comparison of the piperazine sulfonyl moiety positioning in the RT complexes with 18b1 and 25a (PDB 6C0N). Created with PyMOL (http://www.pymol.org/) and with BioRender.com.

Full size image

Molecular dynamics (MD) simulation study

Subsequently, the MD simulation was conducted to further explain the differences in the inhibitory activity of 18b1 and ETR against RT variants. Three x-ray structures of HIV-1 RTs3, including K103N RT (PDB code: 6C0O), E138K RT (PDB code: 6C0P) and V106A/F227L RT (PDB code: 6DUF) were downloaded from RCSB PDB (rcsb.org)38.

Figure 6 shows the root mean square deviation (RMSD) of 18b1 and ETR during 300 ns MD simulation with corresponding RT variants. It is obvious that 18b1 has clustered into distinct conformations, which is clear from the pattern and the range of RMSD values. On the contrary, the RMSD plot for ETR shows a stable binding conformation with protein. The most dominant cluster analysis of 18b1 with corresponding RT variants shows that they have similar binding modes (Fig. 7a, c, e). However, the conformation of piperazinyl and the orientation of methyl sulfone are varied. The cyano group of 18b1 is orientated towards the left wing, except K103N variant, which suggest the unique orientation for the cyano group is caused by the mutation of K103 to N103 when binding this variant. The binding modes of representative structure of the most dominant clusters of 18b1 and ETR bound to the three RT variants are similar regarding the orientation of right and left wings of both compounds in all RT variants binding (Fig. 7).

Fig. 6: RMSD (heavy atoms) in reference to the first frame during the 300 ns MD simulation.
figure 6

a K103N-18b1, b K103N-ETR, c E138K-18b1, d E138-ETR, e V106A/F227L-18b1 and f V106A/F227L-ETR.

Full size image
Fig. 7: Interactions in the representative structures (most abundant cluster) of NNRTIs and HIV-1 RT.
figure 7

a K103N-18b1, b K103N-ETR, c E138K-18b1, d E138-ETR, e V106A/F227L-18b1 and f V106A/F227L-ETR. Mutated amino acids are presented inside a box. Amino acids that are not involved in interactions with ETR are represented in lines. Inhibitor is represented as balls and sticks.

Full size image

Figure 7 shows the binding poses of 18b1 and ETR to the binding sites of the K103N, E138K and V106A/F227L RT variants in the most abundant cluster of each binding. Supplementary Table S2 shows detailed interactions, hydrogen bonding and hydrophobic interaction, of both compounds to the three RT variants. The general view of Fig. 7 and Supplementary Table S2 shows that 18b1 forms more interactions with the surrounding amino acids comparing ETR.

Hydrogen bond interactions between 18b1 and ETR and corresponding RT variants were investigated by hydrogen bond analysis implemented in cpptraj (Supplementary Table S2). Supplementary Table S2 shows that the NH group on the right wing of 18b1 and ETR form hydrogen bonds with the backbone oxygen of K103 (N103 in K103N). Interestingly, in contrast to the interactions with other variants, both inhibitors contact the backbone HN of K101 in E138K variant. This could be explained by the lost hydrogen bond between K101 and K138 in E138K, as presented in the other RT variants according to the hydrogen bond analysis. This made K101 orientates to a position that can interact with the inhibitor more frequently than other variants.

The other type of hydrogen bond takes place between C-H and oxygen or nitrogen atom39,40,41,42. It is considered an important bonding force in biomolecules, despite it is a weak interaction. Supplementary Table S2 shows that the oxygen atoms of 18b1 sulfone (O1 and O2) are involved in aliphatic hydrogen bonding with residues F227, L234 and P236 at good frequency. In K103N variant there are interactions to P225, F227, and P236 with sulfone oxygen atoms. Also, it establishes conventional hydrogen bonds with N103, which is not present in the binding of ETR. The interactions of sulfone oxygen atoms with surrounding amino acids added extra bonding forces, which were reflected on the activity of 18b1 against each RT variant. Interestingly, the piperazine moiety of 18b1 could forms additional interactions with the backbone atoms (H and O) of L234, H235 and P236 in the three variants frequently (Supplementary Table S2). These interactions are rarely found in the binding of ETR. Forming hydrogen bonds with the backbone atoms has advantage over interactions with the side chain of amino acids since the mutation will have minimal effect on the activity of these inhibitors. This could explain the activity of 18b1 against the mutant variants compared to the activity of ETR.

Hydrophobic interactions were also investigated in Supplementary Table S2. It is clear that the hydrophobic interactions are dominant when binding to RT variants. 18b1 shows interactions with P95 in E138K, whereas it is not present in the interaction of ETR to any RT variant. Interactions with E138 are present between 18b1 and K103N. Also, 18b1 interacts with P236 in all variants. These extra hydrophobic interactions and the abovementioned hydrogen bonds suggest 18b1 behaves better potency comparing ETR, which is consistent with its experimental activity.

The binding free energies of 18b1 and ETR with corresponding RT variants were calculated using MMPBSA43 and were represented in Supplementary Table S3. The Van der Waals (vdw) contribution for 18b1 bound to the variants is higher (more negative Evdw) than the contribution of ETR, which shows the importance of hydrophobic interactions in their activities. Also, it is clear that the contribution to V106A/F227L is the highest followed by E138K and K103N. The nonpolar contribution to the solvation energy shows favorable binding of both 18b1 and ETR with a clear preference to the binding of 18b1. The electrostatic contribution was the highest for K103N followed by V106A/F227L and E138K. The electrostatic contribution to the solvation free energy was the highest for V106A/F227L, followed by K103N and E138K, which shows effect on the binding of 18b1 to each variant. Supplementary Fig. S3 shows the free energy decomposition of binding site amino acids, which elucidates amino acid binding energies, individually (vdw, electrostatic, polar solvation and nonpolar solvation energies). The electrostatic contribution of E138 favors the binding of both 18b1 and ETR to all variants except E138K. According to Supplementary Fig. S3, vdw contribution of E138 favors the binding of 18b1 in all variants, except in E138K which slightly favors the binding of ETR. The mutation of E138 to K138 made the contribution of the polar solvation energy favored, where it is negative in E138K and positive in other variants. Inspecting the vdw interactions in Supplementary Fig. S3 shows that the majority of amino acids favor the binding of 18b1 to all RT variants. Also, the electrostatic energy of most amino acids favors the binding of 18b1 to all RT variants. Furthermore, the nonpolar salvation energy is more negative in the binding of 18b1 to all RT variants.

Fsp3 value, water solubility, milogP and ligand efficiency

“Fraction of sp3 carbon atoms” (Fsp3) is used to characterize the carbon saturation of molecules, and it is considered an important parameter for drug-like properties44,45. According to the statistics, about 84% of approved drugs possess Fsp3 values above 0.4219,41. As indicated in Table 4, ETR suffers from a low Fsp3 value (0.10) due to the presence of multiple aromatic fragments, which led to poor water solubility (<1 μg/mL at pH 7.0) and undesirable milogP (5.03). By introduction of piperazine methylsulfonyl, 18b1 possessed acceptable water solubility (13.46 μg/mL at pH 7.0) and reasonable milogP (4.52) with an improved Fsp3 content (0.27). Furthermore, ligand efficiency (LE) as an emerging index is used to evaluate the balance between biological activity and drug-likeness46,47. A promising lead compound is considered to have an LE value above 0.348,49, so 18b1 displayed appropriate LE value (0.32) as with ETR (0.41).

Table 4 Fsp3, water solubility, milogP and LE values of 18b1 and ETR.
Full size table

In vitro cytochrome P450 (CYP) enzymatic inhibitory activity

It is reported that ETR and RPV inhibit CYP enzymes, which could cause potential drug-drug interactions and limit the co-administration of multiple drugs50. Therefore, 18b1 was assessed for its inhibitory potency against the five main CYP isozymes. ETR and selected inhibitors were used as control. As shown in Table 5, 18b1 exhibited weak inhibitory activity against CYP2C9 (IC50 = 5.30 μM) and CYP2C19 (IC50 = 8.00 μM), while conversely ETR (IC50 = 0.277 μM and 0.496 μM, respectively) and RPV (IC50 = 0.346 μM and 0.335 μM, respectively) were submicromolar CYP2C inhibitors. On the other hand, 18b1 and ETR demonstrated no obvious inhibitory effect (>10 μM) towards CYP1A2, CYP2D6 and CYP3A4. These results indicate that 18b1 is less likely to cause CYP-mediated drug-drug interactions.

Table 5 Effects of 18b1 on inhibition of CYP1A2, CYP2C9, CYP2C19, CYP2D6 and CYP3A4.
Full size table

In vivo pharmacokinetics study

Compound 18b1 was further selected to evaluate its pharmacokinetics (PK) profile in Sprague-Dawley (SD) rat model. As described in Table 6, 18b1 demonstrated acceptable terminal half-life (T1/2 = 2.42 h), moderate clearance (CL = 2.54 L/h/kg) and favorable distribution volume (V = 8.81 L/kg) after an intravenous dose of 2.0 mg/kg. When orally administered at a dose of 10.0 mg/kg, the maximum concentration (Tmax) of 18b1 was 16.05 ng/mL at 3.50 h (Tmax) and the T1/2 was 3.05 h. Nevertheless, the oral bioavailability (F) of 18b1 was detected to be 1.34%, which requires further optimization, though this is more serious for ETR (undetectable51).

Table 6 Pharmacokinetics profiles of 18b1a.
Full size table

Discussion

In this research, compound BH-11c was utilized as a lead from which we designed a series of DAPY-typed NNRTIs, aiming at enhancing backbone-binding interactions with the residues in tolerant region I. Interestingly, the newly synthesized compound 18b1 demonstrated significantly improved antiretroviral activities compared to BH-11c against all the tested HIV-1 strains. Furthermore, 18b1 possessed single-digit nanomolar potency against the wild-type and five mutant HIV-1 strains, including L100I, K103N, Y181C, E138K and F227L/V106A. The co-crystal structure indicated that the sulfonyl group of 18b1 formed hydrogen bonds with main-chain atoms of F227 and L234, and the cyano group displayed hydrophobic contacts with side-chain atoms of V106 and F227. Further MD simulation studies were conducted to explain the differences in the inhibitory activity of 18b1 and ETR against RT variants. Compared to ETR (<1 μg/mL at pH 7.0), 18b1 displayed improved water solubility (13.46 μg/mL at pH 7.0) with an appropriate LE value (0.32). Moreover, 18b1 revealed significantly lower inhibitory activity than ETR and RPV against CYP2C9 and CYP2C19, indicating that 18b1 was less likely to cause drug-drug interactions. Nevertheless, 18b1 requires further optimization to improve the oral bioavailability (F = 1.34%). Consequently, we consider 18b1 as a promising lead compound worthy of further study.

Methods

Molecular simulation methods

See Section 1 in Supplementary Information.

Synthetic procedures

See Section 2 in Supplementary Information. For original spectra of compounds see Supplementary Data 2.

Anti-HIV activity test studies

See Section 3 in Supplementary Information.

HIV-1 RT inhibition assays

See Section 4 in Supplementary Information.

HIV-1 RT crystallization and structure determination (including Supplementary Figs. S1–2, Supplementary Table S1)

See Section 5 in Supplementary Information. For PDB file see Supplementary Data 1.

Molecular dynamics simulation methods (including Supplementary Fig. S3, Supplementary Tables S2–3)

See Section 6 in Supplementary Information.

Water solubility measurements

See Section 7 in Supplementary Information.

CYP enzyme inhibition assay

See Section 8 in Supplementary Information.

Pharmacokinetics assays

See Section 9 in Supplementary Information.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.